# Resetting of centre of mass velocity

Dear Users,

I am solving a convective heat transfer type problem using Lammps.

I have generated flow in the positive X direction using addforce command. And the thermostat was done by removing center of mass velocity.

For finding the velocity profile at a section, I have divided into small regions at that section and I calculated center of mass velocity at each sub region. At the outlet section (positive X axis end) the velocity profile was found as parabolic. I have given periodic boundary condition to the X direction. So the velocity profile at the inlet (negative X axis end) section will be parabolic profile because of periodic boundary condition.

I would like to maintain uniform velocity profile at the inlet section. Is there is any way of doing the same?

Any help would be appreciated. Thanks very much!

Dear Users,

I am solving a convective heat transfer type problem using Lammps.

I have generated flow in the positive X direction using addforce command.
And the thermostat was done by removing center of mass velocity.

For finding the velocity profile at a section, I have divided into small
regions at that section and I calculated center of mass velocity at each sub
region. At the outlet section (positive X axis end) the velocity profile
was found as parabolic. I have given periodic boundary condition to the X
direction. So the velocity profile at the inlet (negative X axis end)
section will be parabolic profile because of periodic boundary condition.

I would like to maintain uniform velocity profile at the inlet section. Is
there is any way of doing the same?

this is all too vague a description to give any specific advice.

overall, the situation is pretty simple. you define the physics of the
model through your input. if you don't get what you expect, that means
implementing what you define to be your model.

since there are many different ways to set up a simulation with the
same basic parameters as you describe and without knowing the geometry
and dimensions, it is not possible to say which is the case here.

axel.

Dear Sir,

Sorry about the less description. I am simulating argon flow around a carbon nanotube. My simulation box is a cube of 60 angstrom. I have placed CNT at the center of the simulation box. And I applied a spring force to the CNT. Before the inlet section I have created two artificial regions, one is for applying force and second one is for resetting temperature by removing center of mass velocity. By creating these two regions I am able to make constant temperature with uniform flow at the inlet section.

When the argon is crossing the CNT, the velocity profile becomes parabolic as expected. When I am running it for long time the the same argon fluid is entering to the inlet region again due to the periodic condition and because of this the velocity profile at the inlet is changing from uniform to parabolic profile. But I have to maintain it as uniform profile at every timestep. How can I solve this problem?

This has been discussed previously on the mailing list. The system you are trying to model is not periodic, but it is convenient to use a periodic boundary in MD, just so you can recycle the particles, rather than trying to constantly create and destroy. But you need to make sure that the inlet particles have no memory of their past history. See the thread from 11/11/2015 with the subject “Inlet/Outlet non-periodic conditions” for more details.

Aidan

Dear Sir,

Thanks very much for your responses. I think that thread will be sufficient for debugging my problem. I will try these out and let you know.

Dear LAMMPS developers,
In the repository lib/meam/ is a new .c file (fm_exp.c) that the Makefile.ifort does not take into account and hence generate compiling errors.
Some lines have to be added as:

OBJ = \(SRC:\.F=\.o\) fm\_exp\.o CC = icc %\.o:%\.c &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;(CC) -c \$<

quite as in the Makefile.gfortran which has been well changed.
Is it possible to add those lines in the last svn version?

Thank you very much,

Best regards.

Antoine Jay

Dear LAMMPS developers,
In the repository lib/meam/ is a new .c file (fm_exp.c) that the Makefile.ifort does not take into account and hence generate compiling errors.
Some lines have to be added as:

OBJ = \(SRC:\.F=\.o\) fm\_exp\.o CC = icc %\.o:%\.c &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;(CC) -c \$<

quite as in the Makefile.gfortran which has been well changed.
Is it possible to add those lines in the last svn version?

thanks for your report. i've added those changes (and to a couple more
makefiles) to my development branch and well forward those to steve as
soon as possible.

axel.

Dear Sir,

I have tried with append/atoms command, but the problem is that it runs only in non-periodic boundary condition.

But I need to recycle the particles from outlet to inlet without the memory of their past center of mass velocity. For this I have created one extra region at the outlet end and I applied langevin thermostat, but still inlet particles having their past history. How can I resolve this issue.

Thanks very much!

Recognize that this is not a standard procedure in LAMMPS, so you are going to have to invest considerable effort constructing and validating the simulation. This is not plug and play. And it is not something that we can explain to you on the mailing list, except in a very superficial way. The key issue with the Langevin thermostat is to apply it only to particles in a single region, and to make it strong enough to erase the previous flow profile.

Aidan